Georg Theiner has made available a draft copy of his intro to a special issue (Writing in Mind) “Language, Literacy, and Media Theory: Exploring the Cultural History of the Extended Mind“. I notice that he references Marshall McLuhan whose famous slogan “the medium is the message” informs the start-up company that I’m the co-founder of. Georg is also a very good stigmergy theorist and will be contributing to Ted and my Human Stigmergy: Theoretical Developments and New Applications. Check out Georg’s book below.
A new paper by Giuggioli et al. in PNAS September 30, 2013
Collective animal behavior studies have led the way in developing models that account for a large number of individuals, but mostly have considered situations in which alignment and attraction play a key role, such as in schooling and flocking. By quantifying how animals react to one another’s presence, when interaction is via conspecific avoidance rather than alignment or attraction, we present a mechanistic insight that enables us to link individual behavior and space use patterns. As animals respond to both current and past positions of their neighbors, the assumption that the relative location of individuals is statistically and history independent is not tenable, underscoring the limitations of traditional space use studies. We move beyond that assumption by constructing a framework to analyze spatial segregation of mobile animals when neighbor proximity may elicit a retreat, and by linking conspecific encounter rate to history-dependent avoidance behavior. Our approach rests on the knowledge that animals communicate by modifying the environment in which they live, providing a method to analyze social cohesion as stigmergy, a form of mediated animal–animal interaction. By considering a population of animals that mark the terrain as they move, we predict how the spatiotemporal patterns that emerge depend on the degree of stigmergy of the interaction processes. We find in particular that nonlocal decision rules may generate a nonmonotonic dependence of the animal encounter rate as a function of the tendency to retreat from locations recently visited by other conspecifics, which has fundamental implications for epidemic disease spread and animal sociality.
The delayed response between mark deposition, the action of an individual, and conspecific retreat, the reaction of another conspecific, is a basic ingredient of stigmergy (10, 11), a mediated interaction mechanism whereby the changes produced endogenously in the environment by the marks of one individual elicit a response in the neighbors, which in turn respond, affecting their nearest neighbors. This cascade of events creates a feedback mechanism for the entire population, which self-organizes into a dynamic spatiotemporal pattern.
The shrinking and growth of Si are controlled, respectively, by the aging of the marks and the movement of the animals. The transience of the deposited cues tends to reduce the size of a marked area, because inactive marks are ignored by conspecifics. An aging mark at a given location reduces the propensity of other conspecifics to retreat from that location, which in turn increases the pressure onto individual i to move further inside its own marked area, reducing even further the spatial extent of Si. Because a decrease in the size of Si further reduces its spatial extent, the decay of the marks acts as a positive feedback. The other positive feedback is the movement of the animals, which helps the growth of Si. As animals deposit marks by exploring regions beyond their inner-core areas, they increase the extent of Si and pressure neighbors into moving away to avoid confrontation. This in turn allows them to explore even larger areas, thus further increasing the size of their marked areas. Positive feedback mechanisms act to reinforce a given process and are the key to explaining various forms of aggregation and pattern formation (see, e.g., ref. 12 for the application of reinforced random walks to represent some types of positive feedback). On the other hand, the negative feedback acts in the opposite direction of the variation of Si whether it is a decrease or increase in its spatial extent. As marked areas get smaller, animals may traverse them quicker and thus slow their shrinking. Similarly, as marked areas get bigger, animals take longer to move across them, preventing individuals from re-marking aging marks. This results in a reduction of the growth rate of Si.
We choose to interpret the space use of marking animals as a stigmergic interaction for three reasons. The first is that animal marking is a widespread behavior in the animal kingdom, and although each species has evolved specialized means of communication by depositing cues on the terrain, it serves the general function of broadcasting an animal’s presence. Marks contain information about identity and relative dominance (13), with many vertebrates (14) and eusocial insects (15) making use of chemical signals but also with examples in which visual marks are used, such as feathers and feces by birds. Stigmergy represents a well-developed concept that would help in studying animal space use from a general theoretical perspective, independent of the types of signals present in the marks that get deposited or the sensory modalities required for the detection of those signals. The second reason is that stigmergy makes interactive processes history dependent, which captures the fact—often neglected in quantitative analyses of animal space use—that individuals do not respond simply to the current position of other conspecifics, but also to where they have been in the recent past. A mark, when detected, represents a record of an individual’s past activity in a specific location to which other conspecifics eventually respond. The third reason is that stigmergic stimulus–response association relies upon modification of the environment. As environmental heterogeneity may also affect how individuals move in space, our approach yields a method to quantify another form of spatial heterogeneity, the one generated endogenously from animal interactions. It thus may be possible to extend our current framework to provide a common currency to interpret animal space use as a function of the most important endogenous and exogenous features of the ecosystem, respectively, conspecific avoidance and environmental covariates. Promising approaches in that respect already are available and may help link population spatial distribution to animal spatial memory and landscape persistence (16), as well as to prey distribution and terrain steepness (17, 18).
In this framework of socially interacting animals, we are interested in determining how the individual movement response to the presence of conspecifics shapes the degree of segregation in the population. A useful tool to characterize the emerging spatiotemporal pattern of the population is the encounter rate of mobile animals, an instrument of broad ecological applicability (19). Most encounter estimates have relied upon basic animal movement models, in which displacement is ballistic and individuals are completely independent, which amounts to considering animals as “ideal gas” particles. This approach has been taken as a null model to estimate the frequency of meeting or associations among mobile animals (20) and has been used recently to estimate, with the help of allometric considerations in a spatially implicit context, how home range size scales with body mass (21). Here, to capture the key biological features of the movement and interaction processes that underlie animal spacing, we consider a spatially explicit scenario to determine how individual behavior affects animal space use. The focus of our analysis is the quantification of the average encounter rate, home range size, and degree of exclusivity as a function of the degree of stigmergy.
Fig. 1. Schematic representation of stigmergy in marking animals. When the animal i detects the presence of active foreign marks, it responds by retreating from the locations of the foreign marks (Ri). At the same time, the animal i itself deposits marks over the terrain, whose active set constitutes the stimulus Si that another member j of the population detects, inducing the response Rj. In turn, animal j deposits its own marks (Sj), whose locations affect animal i again or individual k, which will react and itself produce a stimulus. The number of individuals involved in this feedback loop may be as a large as the entire population or as small as just individuals i and j, depending on the locations the animals visit after their response. The dashed lines around Rk and Sk represent the fact that the number of steps necessary to affect individual i may vary because of the random nature of the movement process and, thus, of the probability of animal i encountering the stimulus (Sj).
Here is the intro and conclusion to Chris and my paper:
To know is to cognize, to cognize is to be a culturally bounded, rationality-bounded and environmentally located agent. Knowledge and cognition are thus dual aspects of human sociality. If social epistemology has the formation, acquisition, mediation, transmission and dissemination of knowledge in complex communities of knowers as its subject matter, then its third party character is essentially stigmergic. In its most generic formulation, stigmergy is the phenomenon of indirect communication mediated by modifications of the environment. Extending this notion one might conceive of stigmergy as the extra-cranial analog of artificial neural networks or the extended mind. With its emphasis on coordination, it acts as the binding agent for the epistemic and the cognitive. Coordination is, as David Kirsh (2006, p. 250) puts it, “the glue of distributed cognition”. This paper, therefore, recommends a stigmergic framework for social epistemology to account for the supposed tension between individual action, wants and beliefs and the social corpora: paradoxes associated with complexity and unintended consequences. A corollary to stigmergic epistemology is stigmergic cognition, again running on the idea that modifiable environmental considerations need to be factored into cognitive abilities. In this sense, we take the extended mind thesis to be essentially stigmergic in character.
This paper proceeds as follows. In Section 2, we set out the formal specifications of stigmergy. In Section 3, we illustrate the essentially stigmergic characteristics of social epistemology. In Section 4, we examine extended mind externalism as the preeminent species of stigmergic cognition. In Section 5 we illustrate how the particle swarm optimization (PSO) algorithm for the optimization of a function could be understood as a useful tool for different processes of social cognition, ranging from the learning of publicly available knowledge by an individual knower, to the evolution of scientific knowledge. In Section 6, we offer some concluding remarks.
A great deal of ground has been covered in the course of which we have made a case for two central claims:
1. Social epistemology has the formation, acquisition, mediation, transmission and dissemination of knowledge in complex communities of knowers as its subject matter. Such knowledge is, for the most part, third party and as such it is knowledge that is conditioned and modified. Understood thus, social epistemology is essentially stigmergic.
2. One might conceive of social connectionism as the extra-cranial analog of an artificial neural network providing epistemic structure. The extended mind thesis (at least the Clarkean variant) runs on the idea that modifiable environmental considerations need to be factored into cognitive abilities. This notion of cognition is thus essentially stigmergic.
With 1 and 2 in mind, two disclaimers are in order. First, a stigmergical socio-cognitive view of knowledge and mind should not be construed as (a) the claim that mental states are somewhere other than in the head or, (b) the corollary, that as individualists, we do not think that what is outside the head has nothing to do with what ends up in the head. A stigmergic approach, necessarily dual aspect, does not require one to dispense with one or the other. There is no methodological profit whatsoever to throwing out the Cartesian baby along with the bath water. Second, a socio-cognitive view of mind and knowledge be not be mistaken as a thesis for strong social constructivism, the idea all facts are socially constructed (a denial that reality in some way impinges upon mind) – again, it would be inconsistent with the environmental emphasis entailed by stigmergy.
For Clark, “[M]uch of what goes on in the complex world of humans, may thus, somewhat surprisingly, be understood in terms of so-called stigmergic algorithms.” (Clark, 1996, p. 279). Traditional cases of stigmergic systems include stock markets, economies, traffic patterns, supply logistics and resource allocation (Hadeli, Valckenaers, Kollingbaum, & Van Brussel, 2004), urban sprawl, and cultural memes. New forms of stigmergy have been exponentially expanded through the affordances of digital technology: we’ve expounded upon Google’s RP and Amazon’s CF but of course include wiki, open source software, weblogs, and a whole range of “social media” that comprise the World Wide Web. These particular examples serve to make the wider stigmergical point that the Janus-like aspect of knowledge and cognition must be set against a background fabric of cultural possibility: individuals draw their self-understanding from what is conceptually to hand in historically specific societies or civilizations, a preexisting complex web of linguistic, technological, social, political and institutional constraints.
It is no surprise then that it has been claimed that stigmergic systems are so ubiquitous a feature of human sociality, it would be more difficult to find institutions that are not stigmergic ( Parunak, 2005 and Tummolini and Castelfrananchi, 2007). If stigmergy were merely coextensive with “the use of external structures to control, prompt, and coordinate individual actions” (Clark, 1997, p. 186), then the concept would amount to a claim about situated cognition in all its dimensionality Solomon, 2006b. While stigmergy includes these aspects, it distinctively emphasizes the cybernetic loop of agent → environment → agent → enviro nment through an ongoing and mutual process of modification and conditioning, appearing to dissolve the supposed tension between the self-serving individual and the social corpora at large through indirect interaction. Though this process of behavior modification has long since been identified by both PSE and SSE theorists, only recently has there begun a concerted effort ( Turner, 2001 and Turner, 2003) to, as Ron Sun puts it (Sun, 2006) “cognitivize” human sociality. Social theory and cognitive science must now recognize the virtues of a “cognitivized” approach to all things social.
This from Wired.
During the 130 million years or so that ants have been around, evolution has tuned ant colony algorithms to deal with the variability and constraints set by specific environments.
Ant colonies use dynamic networks of brief interactions to adjust to changing conditions. No individual ant knows what’s going on. Each ant just keeps track of its recent experience meeting other ants, either in one-on-one encounters when ants touch antennae, or when an ant encounters a chemical deposited by another.
Such networks have made possible the phenomenal diversity and abundance of more than 11,000 ant species in every conceivable habitat on Earth. So Anternet, and other ant networks, have a lot to teach us. Ant protocols may suggest ways to build our own information networks…
Speaking of what we can learn from ants, here is a project that I’m co-editing.
Eliot’s intro and first section to his paper:
Human cognition mostly takes place in the context of other people. This is true in two ways. First, if we consider the immediate context of other people who are physically present, they may influence or even help constitute an individual’s cognition by providing information, agreeing or disagreeing, being part of a group decision-making process, etc. (Tollefsen, 2006 and Wegner, 1986). And as a broader context, the group memberships and socially defined identities that make each of us who we are (e.g. an American, a professor, a father) both motivate and potentially bias our cognition as we move through our lives. As Clancey (1997, p. 366) put it, the “overarching content of thought is not…[descriptions or symbolic representations of states of the world], but coordination of an identity” in a social context. If I sit alone in my office working on a paper for publication, my actions are nevertheless socially shaped, for they ultimately reflect socially defined identities and goals (e.g. to write an interesting paper; to win the approval of professional colleagues; to be a successful researcher; to earn a living for myself and my family). Indeed, a pure case of individual (nonsocial) cognition – cognition that is independent not only from immediate social influences but also from the individual’s network of social relationships, group memberships, and self-identities – is difficult to even imagine.
The field of social psychology has as its defining focus such social influences on individual cognition, affect, and behavior, in both forms (the immediate social context, and the larger web of relationships and identities that shape the individual). Thus, this special issue on situated/embodied/distributed perspectives on social cognition addresses issues that are central to the field of social psychology. For this reason it is interesting to note that these emerging perspectives have actually been introduced to the field only recently (e.g. Barsalou et al., 2003, Semin and Smith, 2002, Semin and Smith, in press and Smith and Semin, 2004) – as much as a decade or two after they were initially advanced within artificial intelligence and cognitive science (Brooks, 1986/1999, Clancey, 1997 and Clark, 1997). However, as argued elsewhere in more detail (Smith & Semin, 2004), despite its recent onset, the integration of situated/embodied/distributed perspectives with the substantive concerns of social psychology is likely to be extraordinarily fruitful, even revolutionary in many respects. The reason is that the merger of these new perspectives, which have mostly been applied to improve our understanding of individual cognition and adaptive behavior, and the emphasis of social psychology on the centrality of the social context of behavior, opens up new vistas for conceptual and theoretical exploration.
This article addresses the intersection of embodied and distributed cognition, a focus that holds special interest from the viewpoint of social psychology. We can conceptualize this intersection in three ways. The first point is simply what these perspectives have in common: both seek to extend our conception of cognition beyond information processing performed by the brain, to include the body and sensory-motor systems (embodied cognition) as well as other bodies and minds (distributed cognition). Second, the principle of embodiment has to date been applied mostly to understanding individual functioning (e.g. the role of motor representations in language comprehension). Adding a distributed cognition perspective suggests that embodiment also has implications beyond the level of the individual, for example with regard to interpersonal cooperation or relationships. Third, socially distributed cognition, such as group problem-solving, has mostly been conceptualized as involving abstract, amodal information processing. But adding the embodiment perspective calls attention to potential embodied influences on group interaction and collective cognition. In fact, it can be argued that an important function of embodiment is to externalize cognitive processes so they can influence and be influenced by others. For example, if someone looks puzzled and scratches his head when trying without success to solve a puzzle or retrieve some information from memory, it may cue others to jump in and offer suggestions or help. If cognition was disembodied – implemented purely by inner computation processes lacking any external signs – distributing cognition across a group of people would be much more difficult.
This paper will discuss two areas within the intersection of the embodiment principle and distributed cognition. First, there are embodied aspects of social relationships as well as of individual-level cognition, and some preliminary evidence is now available on this point. Second, we will examine some general properties of socially distributed cognition (e.g. group problem-solving) in comparison to individual-level cognition. Research in this area has only begun to examine embodiment effects, but we will suggest some relevant possibilities.
1. Embodiment of social relationships
The principle of embodiment has typically been applied in an effort to understand individual-level functioning. For example, research addresses how the physical properties of muscles and limbs ease demands for neural control in locomotion (e.g. Thelen & Smith, 1994) or how multimodal representations of concepts enable language comprehension (e.g. Barsalou, 1999). A broader look at the embodiment concept includes examination of how aspects of social functioning – specifically, social relationships – are signaled and regulated by embodied cues.
The most directly relevant framework for addressing this topic is the relational models theory developed by Fiske (2004), a cognitive anthropologist. Fiske holds that there are four fundamental types of social relationships. Communal sharing (CS) describes a relationship where people focus on what they have in common and share resources as needed; it is typically found between close kin, and among members of cohesive groups, clans, etc. Authority ranking (AR) describes relationships structured by ordered differences in power or status; they are typically found in workplaces and other hierarchical social institutions, and also in many cases between parents and children. Two other types of relationships are argued to be historically more recent developments, and we will have little to say about these. Equality matching (EM) describes equal sharing or tit-for-tat exchange relationships, and market pricing (MP) involves the exchange of goods using assigned values.
Fiske’s work (2004) includes detailed accounts, supported by anthropological evidence across numerous cultures, of the types of embodied cues that are associated with each of these four relationship types. Specifically, CS relationships are said to be embodied by sharing substances such as food, physical closeness and touch, and synchronized bodily movements; AR relationships are embodied by differences in size or vertical position in space. It is valuable to think of these embodiment hypotheses in terms of Barsalou’s (1999) Perceptual Symbol System model, which holds that conceptual knowledge is represented by abstracted and generalized perceptual experiences that can be simulated (partially re-enacted) in context-sensitive ways. Barsalou’s model goes beyond the idea that we use bodily metaphors for types of social relationships, holding instead that perceptual experiences of physical closeness or synchrony or of differences in size or height partially constitute our concepts of relational closeness or differences in power or authority.
This from Froese, Gershenson, and Rosenblueth.
The extended mind hypothesis has stimulated much interest in cognitive science. However, its core claim, i.e. that the process of cognition can extend beyond the brain via the body and into the environment, has been heavily criticized. A prominent critique of this claim holds that when some part of the world is coupled to a cognitive system this does not necessarily entail that the part is also constitutive of that cognitive system. This critique is known as the “coupling-constitution fallacy”. In this paper we respond to this reductionist challenge by using an evolutionary robotics approach to create a minimal model of two acoustically coupled agents. We demonstrate how the interaction process as a whole has properties that cannot be reduced to the contributions of the isolated agents. We also show that the neural dynamics of the coupled agents has formal properties that are inherently impossible for those neural networks in isolation. By keeping the complexity of the model to an absolute minimum, we are able to illustrate how the coupling-constitution fallacy is in fact based on an inadequate understanding of the constitutive role of nonlinear interactions in dynamical systems theory.
Born on this day in 1899. It’s to analytical (social) epistemology’s (and philosophy of mind’s) impoverishment and shame that Hayek is not that well-known beyond the tiresome caricatures. For all my Hayekana see here. The featured image was very generously given to me by the highly exceptional Walt Weimer.